Prevention 2.0

Prevent the risk of suicide... with a cell phone. This is the project led by psychiatrist Philippe Courtet, who worked with a team of computer scientists from Lirmm to develop the Emma application, designed to detect the risk of suicide.

With over 9,000 deaths every year, France has one of the highest suicide rates in Europe. It is even the leading cause of death among 25-34 year-olds. On a global scale, one person ends his or her life every 40 seconds. A real public health problem, and a challenge for psychiatrists faced with this thorny issue: how to prevent the risk of suicide?

We are at a loss when it comes to predicting suicidal risk," explains Philippe Courtet. Patients never commit suicide in front of their psychiatrist during a consultation. Until now, all we've been able to do is diagnose the risk of suicide a posteriori, when we see a patient and discuss with him or her the moments that preceded the act," stresses the head of the psychiatric emergency department at Montpellier University Hospital.

"Recall bias

With one major obstacle: during this consultation, the patient has little or no recollection of these moments. To avoid this "recollection bias" and better understand the context in which the suicide occurred, specialists lacked "a constant monitoring tool to determine the risk of suicide in real time", stresses Philippe Courtet.

To overcome this problem, the specialist came up with the Emma application, which stands for Ecological mental momentary assesment. This auxiliary tool is needed "to monitor the patient at any given moment, in his or her daily life and environment, and not just during a consultation once a month".

In practice, the application uses two types of evaluation, passive and active. "Passive evaluation analyzes the patient's use of his or her phone: calls, messages, social networks..." explain Jérôme Azé and Sandra Bringay of the Montpellier Computer Science, Robotics and Microelectronics Laboratory. While the application knows nothing about the content of exchanges, simply knowing how a person uses his or her phone is a valuable clue. " If we notice that a patient starts using his phone all night long, or that he no longer contacts his friends, or in general that he changes his habits, it's an indicator that perhaps he's not doing well", stresses the psychiatrist.

Precious clues

For the active part of the assessment, Emma will directly solicit the user via "electronic interviews". The app prompts the patient to answer questions such as "How are you?", "Are you anxious?" or "Do you have dark thoughts?". If the answers suggest distress, the app triggers what's known as the intervention module. This is a personalized safety plan that has been drawn up in advance with the patient and his or her psychiatrist," explains Phi-lippe Courtet. The application offers access to a relaxation and stress management module developed by one of the department's doctors. They can also listen to music, and view pre-selected photos or videos.

As a last resort, the application can suggest that the patient contact pre-selected friends and family. " With this device, we're making use of the protective virtues of social connection in suicide prevention," explains the specialist. Emma can also suggest that the patient call 15 or the VigilanS center, a "recontact program" for people who have attempted suicide. " The aim of this safety plan is to ease the very short time that can lead to a suicide attempt," explains the psychiatrist.

Reaching out

Initially, the application is being tested on 100 patients, who will be monitored for 6 months. This phase will enable us to test how patients use Emma, but also to gather as much data as possible, which will be used by Jérôme Azé and Sandra Bringay's team to create an algorithm that will enable them to fine-tune the application's operation. " We are using artificial intelligence, and more specifically deep-learning, to try and build the most accurate suicide risk prediction algorithm possible, which will be used in a second version of the application", explain the computer scientists.

Supported by the FondaMental foundation, this project has already proved its worth. Initial feedback has been positive," explains Philippe Courtet. Patients report that the app's suggestion to call a loved one has made them realize that people do care about them. With Emma, we suggest reaching out to people rather than waiting for them to reach out to us."